| Grant number: | 18/17678-3 |
| Support Opportunities: | Regular Research Grants |
| Start date: | February 01, 2019 |
| End date: | January 31, 2021 |
| Field of knowledge: | Engineering - Electrical Engineering - Telecommunications |
| Principal Investigator: | Aline de Oliveira Neves Panazio |
| Grantee: | Aline de Oliveira Neves Panazio |
| Host Institution: | Centro de Engenharia, Modelagem e Ciências Sociais Aplicadas (CECS). Universidade Federal do ABC (UFABC). Ministério da Educação (Brasil). Santo André , SP, Brazil |
Abstract
Information theoretic learning(ITL) aims at using common concepts from information theory as entropy and mutual information in the context of adaptive filtering, resulting in new criteria that exploits the statistical information of signals in a more complete manner. Since such criteria exploits the probability density functions and the temporal structure of the signals involved in the filtering process, they present good performances in scenarios where classical criteria and algorithms fail, specially in the context of unsupervised equalization, as in the presence of correlated signals or impulsive noise. To estimate the probability densities, kernel density estimators (KDE) have been employed, such as the Parzen windowing method. In the literature, such process has been done using Gaussian kernels, due to its properties and easy manipulation. In this project, we aim to extend the study of ITL based methods, exploiting the use of other functions as kernels. The interest in such a subject comes from the fact that Gaussian kernels are not optimum KDE. Thus, we intend to implement ITL methods with other kernel functions, analysing the performance of the resulting methods analytically and through simulations. (AU)
| Articles published in Agência FAPESP Newsletter about the research grant: |
| More itemsLess items |
| TITULO |
| Articles published in other media outlets ( ): |
| More itemsLess items |
| VEICULO: TITULO (DATA) |
| VEICULO: TITULO (DATA) |